The importance of the global offset, the sum of femoral and acetabular offset, has been underlined in the literature as a key factor for the functional outcome of total hip arthroplasty (THA). However, the acetabular offset is not defined for bi-plane X-rays, a technology providing 3D measurements of the lower limb and commonly used for patients undergoing THA. The aim of this paper is to introduce a measurement method of the 3D acetabular offset with bi-plane X-rays. Our method combines the use of technical and anatomical coordinate systems. The most appropriate definition will be selected based on the best reliability and measurement error. The consequent reliability of the global offset was also assessed. Twenty-eight patients undergoing primary THA were selected retrospectively. Two operators performed three reconstructions for each patients before and after THA. Intraclass correlation (ICC) and smallest detectable change (SDC) were computed for intra-operator, inter-operator and test–retest conditions for all combinations of technical and anatomical coordinate systems. ICCs were good to excellent. One combination was more reliable than others with a moderate mean SDC of 6.3 mm (4.3–8.7 mm) for the acetabular offset and a moderate mean SDC of 6.2 mm (5.6–6.7 mm) for the global offset. This is similar to the reliability and mean SDC of the femoral offset (4.8 mm) approved for clinical use which indicates that this method of acetabular offset measurement is appropriate. This opens a research avenue to better understand the role of the acetabular offset on THA outcomes, which seems overlooked in the literature.